System Design for the LLM Era: Patterns and principles for production-grade AI architecture
暫譯: LLM時代的系統設計：生產級AI架構的模式與原則

STOP building fragile AI wrappers; START designing resilient AI systems.

Key Features:

- From LLM fundamentals to real-world practicalities

- Patterns and principles for architecting LLM-based systems

- Learn from in-depth case studies

- Decouple from premium models using tiered fallback

- Event-driven architectures for decoupling high-latency agentic workflows

- Cost management approaches

- Security strategies for LLM systems

- Glossary of LLM and AI systems design terminology included

Book Description:

Many companies are trying to turn their small AI experiments into big products, but they lack a good plan.

Engineers need a practical guide to building these new AI systems the right way, so that they can handle scale, won't cost too much to build or operate, and perform reliably.

This book is that guide, combining technical depth with breadth and practicality. Starting from LLM fundamentals, the book details the architectural patterns and design principles needed to build production-grade AI systems. In-depth case studies then show you how to apply them to a range of real-world application scenarios, including AI-native IDEs, adaptive learning platforms, and intelligent search solutions.

The book provides a deep, practical look at the real-world challenges and solutions for building systems with LLMs at their core.

What You Will Learn:

- Architect a complete, production-grade AI-powered system from scratch

- Design and mitigate the unique challenges of LLM APIs, like high latency and cost

- Implement key software engineering patterns like circuit breakers and rate limiting for AI systems

- Choose the right databases and data models for AI applications, including vector search engines

- Build a scalable and resilient system that can handle high load and ensure user privacy

Who this book is for:

This book will be an invaluable learning resource for engineers, architects and leads working with LLMs or looking to integrate LLMs into their existing systems.

Table of Contents

- Atomic Units of LLM Systems

- Core Architectural Patterns for LLM System Design

- Case Study - Designing AI-Native IDEs

- Case Study - Adaptive Learning Platform

- Case Study - AI-powered Search for E-commerce Platforms

- Case Study: AI-Powered Customer Support Agent

- Glossary